- Email: [email protected]
Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment
Journal Pre-proof Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment Stepan Vesely, Christian A. Klöckner, Cameron Brick PII:
S0272-4944(19)30461-X
DOI:
https://doi.org/10.1016/j.jenvp.2019.101362
Reference:
YJEVP 101362
To appear in:
Journal of Environmental Psychology
Received Date: 16 July 2019 Revised Date:
7 October 2019
Accepted Date: 9 October 2019
Please cite this article as: Vesely, S., Klöckner, C.A., Brick, C., Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment, Journal of Environmental Psychology (2019), doi: https://doi.org/10.1016/j.jenvp.2019.101362. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment
Stepan Vesely, Norwegian University of Science and Technology, [email protected] Christian A. Klöckner, Norwegian University of Science and Technology, [email protected] Cameron Brick, Department of Psychology, University of Cambridge & Department of Psychology, Universiteit van Amsterdam, [email protected]
Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment
Abstract Pro-environmental behavior has social signaling value. Previous research suggests that enacting pro-environmental behaviors can signal certain personal characteristics, such as social status and trustworthiness, to others. Using an incentivized experiment, we show that people known to behave pro-environmentally are expected to be more cooperative, are preferred as cooperation partners, and elicit more cooperation from others. The presence of pro-environmental individuals may thus motivate others to exert more effort towards reaching cooperative goals, even in situations where individual and group goals are at odds (i.e., social dilemmas). However, people who behaved pro-environmentally were actually no more cooperative than those performing fewer pro-environmental behaviors.
1. Introduction Climate change mitigation and many of the associated pro-environmental behaviors such as reducing air travel and meat consumption (Dietz, Gardner, Gilligan, Stern, & Vandenbergh, 2009) have become contentious and polarizing subjects in the public discourse (McCright & Dunlap, 2011). As a consequence, people may be reluctant to display or advocate mitigation actions in order to avoid conflict, social sanctions and being perceived negatively by others (cf. Bashir, Lockwood, Chasteen, Nadolny, & Noyes, 2013; Brick, Sherman, & Kim, 2017). Also, people may sort into social relationships with those sharing similar views regarding climate protection, which would allow them to exhibit (or not) pro-environmental behaviors more freely, but this assortative matching could lead to even greater opinion polarization in the long term (Böhm, Pfister, Salway, & Fløttum, 2019). One good strategy on how to start gaining more insight into the processes sketched in the previous paragraph is to examine how people behaving pro-environmentally (henceforth referred to as “environmentalists” for short) are perceived by others, and whether such perceptions can have real consequences in terms of subsequent social interaction. There is now a small but growing literature addressing the complex question of how environmentalists are perceived. What previous research shows is that pro-environmental behaviors and attitudes may be able to convey information about a person’s social status (Brooks & Wilson, 2015; Puska, Kurki, Lähdesmäki, Siltaoja, & Luomala, 2016; Sadalla & Krull, 1995; Skippon, Kinnear, Lloyd, & Stannard, 2016; but see Berger, 2017; Welte & Anastasio, 2010), trustworthiness (Fehrler & Kosfeld, 2013; but see Berger, 2017; Puska et al., 2016), and certain personality traits like conscientiousness, agreeableness, and altruism (Puska et al., 2016; Skippon & Garwood, 2011; Skippon et al., 2016). While existing research shows much promise, it has at least two limitations the present work attempts to address. First, we study not only whether behaving pro-environmentally shapes others’ perceptions of the actor, but also whether it can have an impact on subsequent social interaction. Second, we study whether people take others’ environmentalism into account when making decisions with real (financial) stakes on the line, as opposed to hypothetical decisions or survey responses (cf. Camerer & Hogarth, 1999; Kormos & Gifford, 2014; Klein, Hilbig, & Heck, 2017; Vesely & Klöckner, 2018a). 1
We extend the literature on the signaling function of environmentalism by testing whether pro-environmental behavior is perceived by others to signal cooperativeness, and by studying how people respond to such a signal. We also explore whether environmentalists behave more cooperatively themselves. Pro-environmental behavior can thus be linked to a key domain of social interaction (cooperation) through its function as a social signal. In this view, environmentally relevant decisions are not made in isolation from the broader social context, but drive and are driven by social influence processes. People may sometimes refrain from behaving pro-environmentally if they think this could harm their social image (Brick et al., 2017). However, if pro-environmental behavior can serve as a reliable signal of cooperativeness, applied research could then build on this finding to facilitate pro-environmental behavior. Specifically, environmentalists’ greater perceived cooperativeness and others’ greater willingness to cooperate with them could be leveraged as a co-benefit of and further motivation for behaving pro-environmentally. This possibility echoes what has been suggested by Keohane and Victor (2016) in the context of climate change mitigation negotiations: that countries may contribute to climate change mitigation to build their reputation and to secure future cooperation from others (see also Bain et al., 2016; Brekke & Nyborg, 2008). Thus, if pro-environmental behavior is able to signal cooperative tendencies, this can have important practical implications. Before proceeding to our hypotheses, we shall provide a brief outline of their key theoretical underpinnings. We study cooperation using a standard social dilemma setting (the public goods game) where the group’s material welfare is maximized by joint cooperation by everyone in the group, but at the same time an individual is materially better off not cooperating regardless of what the other group members do (Andreoni, 1988; Dawes, 1980). What motivates our prediction that people engaging in pro-environmental behavior are also more cooperative (see H4 below) is that both cooperative and pro-environmental behaviors are driven by pro-social preferences, such as inequity aversion and norm compliance in case of cooperation (Fehr & Schmidt, 1999; Kimbrough & Vostroknutov, 2016; Krupka & Weber, 2013) and attitudes, values, and norms in case of pro-environmental behavior (Bamberg & Möser, 2007; Klöckner, 2013). We specifically propose that pro-environmental and cooperative behaviors may be linked by virtue of their underlying pro-social motives being correlated within individuals. This is consistent with previous research linking cooperation to pro-environmental behaviors and attitudes (Kaiser & Byrka, 2011; Sussman, Lavallee, & Gifford, 2016; Tarditi, Hahnel, Jeanmonod, Sander, & Brosch, 2018; but see Smith & Bell, 1992), as well as with research on associations between different pro-social motives more generally (Bamberg & Möser, 2007; Gächter, Nosenzo, & Sefton, 2013; Kimbrough & Vostroknutov, 2016; Klöckner, 2013). We next propose that if environmentally-friendly individuals are indeed more cooperative outside the lab, observers will have formed a mental association between environmentalism and cooperation through experience (see H1). Because in social dilemmas, interaction with cooperators rather than free-riders is by design more advantageous (see section 2.3.1 and Andreoni, 1988), we expect that environmentalists, if perceived to be cooperative, will be preferred as cooperation partners (see H2). Finally, provided that environmentalists are believed to be cooperative, others should cooperate with them (see H3) due to well-established reciprocity and inequity aversion motives (Fehr & Schmidt, 1999). Hypotheses. H1: Individuals will expect those who perform more pro-environmental behaviors to be more cooperative in social dilemmas. H2: People who perform more pro2
environmental behaviors will be preferred as cooperation partners in social dilemmas. H3: People who perform relatively more pro-environmental behaviors will elicit more cooperation from others. H4: People who perform relatively more pro-environmental behaviors will be more cooperative in social dilemmas. To test our hypotheses, we invited participants to play four public goods games in the laboratory, select their preferred game partners, and respond to questionnaires, as detailed below. 2. Method 2.1 Participants and sessions Two hundred and eight participants (111 females; mean age = 23.94 years, SD = 5.45) recruited from a subject pool maintained by the Vienna Center for Experimental Economics (VCEE) took part in the study. A priori power calculations indicated that a sample of at least 191 participants was required to detect a small effect (partial R2 = .04) with alpha at .05 (twotailed) and statistical power at .80 (Faul, Erdfelder, Lang, & Buchner, 2007). ORSEE (Greiner, 2015) was used for recruitment and z-Tree (Fischbacher, 2007) for programming. Participants were compensated for their time, mean earnings = 35.3 EUR (including payment for additional unrelated tasks, see section B5 in Appendix B). The design was independently reviewed and approved by VCEE staff, in accordance with University of Vienna regulations. 2.2 Questionnaire Participants responded to 28 questions about their previous pro-environmental behaviors (most of the items were adapted from Kaiser, 1998; see Appendix D). We calculated the total score for each participant, with higher scores indicating enactment of more pro-environmental behaviors (M = 18.03, SD = 3.77, α = .70). Participants also answered socio-demographic questions. We then elicited participants’ risk preferences, using a task adapted from Eckel and Grossman (2002). This served as a filler task obscuring the link between the environmental behavior questionnaire and the social dilemma games. Importantly, we later gave participants the opportunity to condition their selection of partners in one of the social dilemma games on the potential partner’s risk preferences (see section 2.5). 2.3 Public goods games Participants played four standard one-shot linear public goods games (Andreoni, 1988) without feedback. In each game, participants formed anonymous groups of four. In each game, each participant was endowed with 20 tokens and had to decide how to distribute the tokens between a Group project and an Individual project. One token invested in the Individual project would yield one point to the investor and zero points to the other three group members. One token invested in the Group project would yield 0.5 points to all four group members. Thus, investing in the Group project (cooperation) maximized the group’s material welfare, while investing in the Individual project yielded higher material utility to the individual. At the end of the experiment, one of the four games was randomly selected for payment with an exchange rate 10 points = 2 EUR. 2.3.1 Group matching (treatments) In the “Most Environmental Game” condition, participants were informed they would be matched with the three most pro-environmental participants in the session (excluding 3
themselves) based on the pre-experimental questionnaire. In “Least Environmental Game”, participants were informed they would be matched with the three least pro-environmental participants in the session. In “Average Environmental Game”, participants were informed they would be matched with the three participants in the session that were closest to that session’s mean environmentalism score. In “Random Matching Game”, participants were informed they would be matched into groups randomly. The order of the four games was randomized across participants. 2.4 Belief elicitation After each game, participants guessed how many tokens others in their matching group contributed to the Group project in total in that game. Participants were given a 2 EUR bonus for guessing others’ total contribution correctly within ± 3 tokens. 2.5 Partner selection Participants next proceeded to play three repeated public goods games as a part of an unrelated study. Important for the present study, they were given an opportunity to select their preferred partners for one of these games from among other participants in the session. Specifically, each participant was shown a table listing all other participants in the session in random order, with the following information displayed in three columns: each participant’s environmental score, risk seeking score and age. The order in which the three types of scores were displayed in the columns was randomized across participants. Thus, participants were able to condition their choice of partners on two additional plausible criteria besides environmental score (i.e., others’ risk preferences and age, see Kocher, Martinsson, Matzat, & Wollbrant, 2015; Thöni, Tyran, & Wengström, 2012). Participants could then indicate on a 7point scale whether they would like to be matched with each of the other participants present. Participants knew that their stated preferences would with some probability determine with whom they would be matched (see Appendix B). 3. Results Figure 1 displays participants’ contributions to the Group project in the four games and their beliefs of how much a co-player in their matching group contributed.
4
Fig. 1: Own actual and co-player’s expected contributions (means and 95% CIs; maximum possible contribution per game was 20 tokens). In Table 1, we report estimates of two regression models testing H1. In Model 1, we regressed participants’ expectations of how much others in their matching group contributed to the Group project on matching group type dummies, with “Average Environmental Game” serving as the baseline category. Model 2 is the same as Model 1, but “Random Matching Game” serves as the baseline category. All reported tests are two-sided. Table 1 Regression of expected contributions on group type
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Constant Observations No. of clusters R2
Model 1 – Average Environmental Game used as baseline b (95% CI) 1.10 (0.62, 1.59)*** -1.59 (-2.09, -1.08)*** -0.43 (-0.88, 0.01) 9.41 (8.71, 10.12)*** 832 208 .03
Model 2 – Random Matching Game used as baseline b (95% CI) 1.54 (1.01, 2.07)*** -1.15 (-1.67, -0.63)*** 0.43 (-0.01, 0.88) 8.98 (8.29, 9.66)*** 832 208 .03
Notes: *** p < .001. Estimation method is robust linear regression with standard errors clustered at the subject level.
We found clear support for H1: Environmentalists were believed to be more cooperative. Specifically, compared to either control group (Average Environmental Game or Random Matching Game), participants expected co-players who were known to perform the most proenvironmental behaviors in one’s session to contribute more to the Group project, while the reverse was true for co-players who were known to perform the fewest pro-environmental behaviors. 5
The test of H2 is presented in Table 2. As described in section 2.5, for every individual in a session, each participant indicated whether they preferred to be matched with them on a 7point scale. For every individual we averaged the preference ratings received from others. Everyone was thus assigned a “mean preference rating”. Higher values indicated that others wanted to be matched with this person, and lower values indicated that others preferred not to be matched with this person. In Table 2, we regressed participants’ mean preference rating on their environmental score, risk seeking score and age score (i.e., the indicators displayed during the rating exercise). We found strong support for H2: Participants preferred to be matched with others who performed many pro-environmental behaviors. Age and risk preferences had no effect on being preferred as cooperation partner. Table 2 Regression of received preference ratings on players’ characteristics Environmental score Risk score Age score Constant Observations R2
b (95% CI) 0.07 (0.06, 0.08)*** 0.00 (-0.01, 0.01) 0.00 (-0.01, 0.00) 3.12 (2.93, 3.31)*** 208 .51
Notes: *** p < .001. Linear regression with robust standard errors.
Table 3 presents estimates of two regression models simultaneously testing H3 and H4. In Model 3, we regressed participants’ contributions to the Group project on their environmental score (to test H4) and on matching group type dummies (to test H3), with Average Environmental Game serving as the control group. Model 4 is the same as Model 3, but Random Matching Game serves as control. Table 3 Regression of contributions on environmental score and group type
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Environmental score Constant Observations No. of clusters R2
Model 3 – Average Environmental Game used as baseline b (95% CI) 1.22 (0.58, 1.86)*** -1.86 (-2.46, -1.25)*** -0.00 (-0.58, 0.57) 0.11 (-0.09, 0.30) 7.00 (3.33, 10.68)*** 832 208 .03
Model 4 – Random Matching Game used as baseline b (95% CI) 1.23 (0.64, 1.81)*** -1.85 (-2.61, -1.09)*** 0.00 (-0.57, 0.58) 0.11 (-0.09, 0.30) 7.00 (3.32, 10.68)*** 832 208 .03
Notes: *** p < .001. Robust linear regression with standard errors clustered at the subject level.
We found clear support for H3: Environmentalists elicited more cooperation from others. Specifically, compared to either control group, participants contributed more to the Group project when matched with co-players who were known to perform the most proenvironmental behaviors in one’s session, while they contributed less to the Group project when matched with co-players who were known to perform the fewest pro-environmental behaviors. Hypothesis H4 was not supported: Those who reported performing more pro6
environmental behaviors did not cooperate more than those performing fewer proenvironmental behaviors. 4. Discussion and conclusions In this paper we focused on the signaling value of pro-environmental behavior and specifically on its ability to signal cooperative tendencies. We posited that pro-environmental behavior can serve as a signal of cooperation (H1), and that others would act on this signal (H2, H3), and we also proposed that environmentalists would themselves be more cooperative (H4). As predicted, environmentalists were perceived as more cooperative in the social dilemma setting (H1), they were sought after as cooperation partners (H2), and they elicited more cooperation from others (H3). Unexpectedly, environmentalists were not more cooperative compared to those performing fewer pro-environmental behaviors (H4). These findings contribute to the literature on the signaling value of environmentalism (e.g., Berger, 2017; Brick et al., 2017; Griskevicius, Tybur, & Van den Bergh, 2010) by showing that environmental behaviors signal a cooperative disposition and that others act on this signal in situations with real consequences. An important implication of our results is that being seen as an environmentalist confers personal side-benefits to environmentalists, as others are more cooperative towards such individuals, and more willing to establish interactions with them (a form of social capital, see Cinyabuguma, Page, & Putterman, 2005). Such side-benefits could be leveraged by policymakers to promote further pro-environmental action (Bain et al., 2016; Griskevicius et al., 2010; Keohane & Victor, 2016). Environmentalists, however, do not appear to fulfill others’ expectations of being more cooperative themselves (see H4). Subsequent research should therefore explore how to motivate environmentalists to meet others’ initial positive expectations (see e.g. Bellemare, Sebald, & Suetens, 2018). Otherwise, cooperation is likely to break down in repeated encounters when initial positive expectations are not met (Fischbacher & Gächter, 2010). A possible explanation for failing to support H4 could be that cooperative decisions were made anonymously in our experiment, while past cooperative decisions involved in expectation formation had to be publicly observable. Decision observability promotes pro-social behavior (Bradley, Lawrence, & Ferguson, 2018), which may help explain why anonymous decisions deviate from others’ expectations. An intriguing hypothesis consistent with our data is that environmentalists are particularly sensitive to being publicly observed. This suggests possible extensions of the present study, linking it to research on the role that decision observability plays in environmental behavior (Griskevicius et al., 2010; Brick et al., 2017; Vesely & Klöckner, 2018b). A limitation of our study is that environmental behaviors were self-reported, rather than objectively measured (Kormos & Gifford, 2014; Lange & Dewitte, 2019). Another limitation is that we relied on a single decision paradigm, the “give some” public goods game, and replications using other types of social dilemmas could point to possible boundary conditions of the reported effects. There are for example subtle behavioral differences between “give some” and “take some” social dilemmas (Fosgaard, Hansen, & Wengström, 2014). One could speculate that different types of pro-environmental behavior (e.g., investment vs. curtailment) could be better able to signal cooperativeness in specific types of dilemmas (such as the give some vs. take some dilemmas). Another limitation is that only two “distractor” characteristics (age and risk preferences) were used along with environmentalism to inform participants about their potential cooperation partners. Finally, we cannot altogether rule out experimenter 7
demand, as information about others’ pro-environmental behavior was somewhat prominent in the decision tasks and some participants may have felt compelled to react to it for this reason.
References Andreoni, J. (1988). Why free ride? Strategies and learning in public goods experiments. Journal of Public Economics, 37, 291-304. Bain, P. G., Milfont, T. L., Kashima, Y., Bilewicz, M., Doron, G., Garðarsdóttir, R. B., et al. (2016). Co-benefits of addressing climate change can motivate action around the world. Nature Climate Change, 6, 154-157. Bamberg, S., & Möser, G. (2007). Twenty years after Hines, Hungerford, and Tomera: A new meta-analysis of psycho-social determinants of pro-environmental behaviour. Journal of Environmental Psychology, 27, 14-25. Bashir, N. Y., Lockwood, P., Chasteen, A. L., Nadolny, D., & Noyes, I. (2013). The ironic impact of activists: Negative stereotypes reduce social change influence. European Journal of Social Psychology, 43, 614-626. Bellemare, C., Sebald, A., & Suetens, S. (2018). Heterogeneous guilt sensitivities and incentive effects. Experimental Economics, 21, 316-336. Berger, J. (2017). Are luxury brand labels and “green” labels costly signals of social status? An extended replication. PLoS ONE, 12, e0170216. Böhm, G., Pfister, H.-R., Salway, A. & Fløttum, K. (2019). Remembering and communicating climate change narratives: The influence of world views on selective recollection. Frontiers in Psychology, 10, article 1026. Brekke, K. A., & Nyborg, K. (2008). Attracting responsible employees: Green production as labor market screening. Resource and Energy Economics, 30, 509-526. Brick, C., Sherman, D. K., & Kim, H. S. (2017). “Green to be seen” and “brown to keep down”: Visibility moderates the effect of identity on pro-environmental behavior. Journal of Environmental Psychology, 51, 226-238. Brooks, J. S., & Wilson, C. (2015). The influence of contextual cues on the perceived status of consumption-reducing behavior. Ecological Economics, 117, 108-117. Camerer, C., & Hogarth, R. M. (1999). The effects of financial incentives in experiments: A review and capital-labor-production framework. Journal of Risk and Uncertainty, 19, 7-42. Cinyabuguma, M., Page, T., & Putterman, L. (2005). Cooperation under the threat of expulsion in a public goods experiment. Journal of Public Economics, 89, 1421-1435. Dawes, R. M. (1980). Social dilemmas. Annual Review of Psychology, 31, 169-193.
8
Dietz, T., Gardner, G. T., Gilligan, J., Stern, P. C. & Vandenbergh, M. P. (2009). Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Proceedings of the National Academy of Sciences, 106, 18452-18456. Eckel, C., & Grossman, P. J. (2002). Sex differences and statistical stereotyping in attitudes toward financial risk. Evolution and Human Behavior, 23, 281-295. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175-191. Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. The Quarterly Journal of Economics, 114, 817-868. Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10, 171-178. Fischbacher, U., & Gächter, S. (2010). Social preferences, beliefs, and the dynamics of free riding in public good experiments. American Economic Review, 100, 541-556. Fosgaard, T. R., Hansen, L. G., & Wengström, E. (2014). Understanding the nature of cooperation variability. Journal of Public Economics, 120, 134-143. Gächter, S., Nosenzo, D., & Sefton, M. (2013). Peer effects in pro-social behavior: Social norms or social preferences? Journal of the European Economic Association, 11, 548-573. Greiner, B. (2015). Subject pool recruitment procedures: Organizing experiments with ORSEE. Journal of the Economic Science Association, 1, 114-125. Griskevicius, V., Tybur, J. M., & Van den Bergh, B. (2010). Going green to be seen: Status, reputation, and conspicuous conservation. Journal of Personality and Social Psychology, 98, 392-404. Kaiser, F. G. (1998). A general measure of ecological behavior. Journal of Applied Social Psychology, 28, 395-422. Kaiser, F. G., & Byrka, K. (2011). Environmentalism as a trait: Gauging people’s prosocial personality in terms of environmental engagement. International Journal of Psychology, 46, 71-79. Keohane, R. O., & Victor, D. G. (2016). Cooperation and discord in global climate policy. Nature Climate Change, 6, 570-575. Kimbrough, E. O., & Vostroknutov, A. (2016). Norms make preferences social. Journal of the European Economic Association, 14, 603-638. Klein, S. A., Hilbig, B. E., & Heck, D. W. (2017). Which is the greater good? A social dilemma paradigm disentangling environmentalism and cooperation. Journal of Environmental Psychology, 53, 40-49.
9
Klöckner, C. A. (2013). A comprehensive model of the psychology of environmental behavior: A meta-analysis. Global Environmental Change, 23, 1028-1038. Kocher, M. G., Martinsson, P., Matzat, D., & Wollbrant, C. (2015). The role of beliefs, trust, and risk in contributions to a public good. Journal of Economic Psychology, 51, 236-244. Kormos, C., & Gifford, R. (2014). The validity of self-report measures of proenvironmental behavior: A meta-analytic review. Journal of Environmental Psychology, 40, 359-371. Krupka, E. L., & Weber, R. A. (2013). Identifying social norms using coordination games: Why does dictator game sharing vary? Journal of the European Economic Association, 11, 495-524. Lange, F., & Dewitte, S. (2019). Measuring pro-environmental behavior: Review and recommendations. Journal of Environmental Psychology, 63, 92-100. McCright, A. M., & Dunlap, R. E. (2011). The politicization of climate change and polarization in the American public's views of global warming, 2001-2010. The Sociological Quarterly, 52, 155-194. Puska, P., Kurki, S., Lähdesmäki, M., Siltaoja, M., & Luomala, H. (2016). Male-male status signaling through favoring organic foods: Is the signaler perceived and treated as a friend or a foe? Psychology & Marketing, 33, 843-855. Sadalla, E. K., & Krull, J. L. (1995). Self-presentational barriers to resource conservation. Environment and Behavior, 27, 328-353. Skippon, S., & Garwood, M. (2011). Responses to battery electric vehicles: UK consumer attitudes and attributions of symbolic meaning following direct experience to reduce psychological distance. Transportation Research Part D, 16, 525-531. Skippon, S. M., Kinnear, N., Lloyd, L., & Stannard, J. (2016). How experience of use influences mass-market drivers’ willingness to consider a battery electric vehicle: A randomised controlled trial. Transportation Research Part A, 92, 26-42. Smith, J. M., & Bell, P. A. (1992). Environmental concern and cooperative-competitive behavior in a simulated commons dilemma. The Journal of Social Psychology, 132, 461-468. Sussman, R., Lavallee, L. F., & Gifford, R. (2016). Pro-environmental values matter in competitive but not cooperative commons dilemmas. The Journal of Social Psychology, 156, 43-55. Tarditi, C., Hahnel, U. J. J., Jeanmonod, N., Sander, D., & Brosch, T. (2018). Affective dilemmas: The impact of trait affect and state emotion on sustainable consumption decisions in a social dilemma task. Environment and Behavior. Thöni, C., Tyran, J.-R., & Wengström, E. (2012). Microfoundations of social capital. Journal of Public Economics, 96, 635-643.
10
van der Werff, E., Steg, L., & Keizer, K. (2014). Follow the signal: When past proenvironmental actions signal who you are. Journal of Environmental Psychology, 40, 273282. Vesely, S., & Klöckner, C. A. (2018a). Global social norms and environmental behavior. Environment and Behavior, 50, 247-272. Vesely, S., & Klöckner, C. A. (2018b). How anonymity and norms influence costly support for environmental causes. Journal of Environmental Psychology, 58, 27-30. Welte, T. H. L., & Anastasio, P. A. (2010). To conserve or not to conserve: Is status the question? Environment and Behavior, 42, 845-863.
11
Online Appendix A: Descriptive statistics Table A1 presents key descriptive statistics, namely means and SDs (in brackets) summarizing participants’ contributions to the Group project in the four games and their beliefs of how much each of the co-players in their matching group contributed on average. Raw data are available upon request. Table A1 Summary of contributions and beliefs (means, SDs)
Own contribution to the Group project Expected average contribution to the Group project by a co-player
Most Environmental Game 10.16 (6.60)
Least Environmental Game 7.08 (6.24)
Average Environmental Game 8.94 (6.18)
Random Matching Game 8.93 (6.44)
10.52 (5.41)
7.83 (5.10)
9.41 (5.17)
8.98 (5.01)
12
Online Appendix B: Further details on methods B1. Risk elicitation task (see section 2.2) To elicit risk preferences, we used a task where one had to choose one of eight incentivized lotteries (played for real money) involving various degrees of risk (Eckel & Grossman, 2002). An example of a low risk lottery was a prospect of getting either 4.2 EUR or 6.0 EUR with equal probability, whereas an example of a high risk lottery was a prospect of getting either 1.4 EUR or 10.8 EUR with equal probability. Participants were assigned a risk seeking score which was equal to 1 if the safest lottery was selected and equal to 8 if the most risky lottery was selected, with values in between for lotteries involving intermediate levels of risk. B2. Recoding responses when calculating scale reliability (see section 2.2) When calculating scale reliability for the 28-item environmental behavior questionnaire, all values were recoded so that higher values indicated more pro-environmental behavior. We also did the following changes: For item “Are you de-icing the refrigerator regularly?”, the response “I share the refridgerator with other people and someone else is responsible for this” was coded as missing, as it did not unambiguously reflect a pro- or anti-environmental behavior. We acknowledge that this coding decision (and similar coding decisions below) reflect a judgment call on our part. For item “Are you getting your electricity from a provider that guarantees 100% regenerative energy production?”, the response “I can not influence at all from which provider to get electricity” was coded as missing. For item “Do you use a tumble dryer to dry clothes?”, the response “I do not have a tumble dryer” was recoded as “No”. For item “Do you bring your own coffee (tea) cup to school/work (instead of using cups from the coffee machine)?”, the response “I do not drink coffee (tea) at school/work” was coded as missing. For item “Do you drive a car with low fuel consumption or an alternative fuel car (e.g., an electric vehicle)?”, the response “I do not drive” was assigned the same value as “Yes” (i.e., both being coded as the maximally pro-environmental response). For item “Do you drive your car defensively to safe fuel?”, the response “I do not drive” was assigned the same value as “Yes” (i.e., both being coded as the maximally pro-environmental response). For item “Do you return dead batteries to collection points for dangerous waste?”, the response “I do not use batteries” was coded as missing. For item “Do you buy organic meat products?”, the response “I do not eat dairy products” was coded as missing. For item “Do you buy organic meat products?”, the response “I do not eat meat products” was assigned the same value as “Yes” (i.e., both being coded as the maximally pro-environmental response).
13
B3. Comprehension check (see section 2.3) Before proceeding to actual game play, participants had to correctly answer control questions checking their comprehension of the game, and were given an opportunity to try out different investment strategies using an interface automatically calculating outcomes for each selected strategy. This was done in order to ensure that participants fully understood the social dilemma structure of the decision situation. B4. Partner selection: Details of the matching protocol (see section 2.5) Here we describe how we ensured that the preferences for cooperation partners elicited from participants in the task described in section 2.5 in the paper were, with some probability, able to determine with whom participants were subsequently matched. Technically, after participants submitted their preferences, the computer randomly assigned a “position” to every participant between 1 and n, where n was the number of participants in the session. The matching then proceeded in the following way: First, the participant with the highest position was matched with his or her three most preferred co-players (with ties broken at random). Second, the participant with the highest position among the un-matched participants was matched with his or her three most preferred co-players among those that have not yet been matched (again with ties broken at random), and so on until everyone was matched. Participants were made aware of this procedure beforehand. B5. Unrelated tasks conducted in the session (not reported here) Participants also played an additional one-shot public goods game and three iterated public goods games at the end of the session.
14
Online Appendix C: Additional analyses C1. Non-parametric tests of H1 and H3 Results of Wilcoxon signed-rank tests lead to the same conclusion as the regression tests of H1 reported in Table 1 in the paper. Specifically, expected contributions by matching group members were higher in Most Environmental Game than in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). Expected contributions by matching group members in Least Environmental Game were, as predicted, lower than in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). Results of Wilcoxon signed-rank tests similarly lead to the same conclusion as the regression tests of H3 reported in Table 3 in the main text. Contributions to the Group project were higher in Most Environmental Game than contributions in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). Contributions in Least Environmental Game were, again as predicted, lower than in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). C2. Robustness (exploratory) analyses: Prediction of expected contributions In Table C1 we repeat analyses reported in Table 1 in the paper, with additional controls (own environmental score, risk preferences, age and gender) added to the models. In Model C1, we regressed participants’ expectations of how much others in their matching group contributed to the Group project on matching group type dummies and the additional control variables. “Average Environmental Game” matching group served as the baseline category. Model C2 is the same as Model C1, but the “Random Matching Game” group served as the baseline category. As in the main analyses, hypothesis H1 received support also in the robustness analyses (in fact, matching group dummy coefficients remain virtually unchanged). Interestingly, individuals performing more pro-environmental behaviors expected others to contribute less. Older participants expected others to contribute more. The latter results should be interpreted with caution, as these analyses are exploratory in nature. Table C1 Regression of expected contributions on matching group type and additional controls
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Environmental score Risk preferences Age Female Constant Observations No. of clusters R2
Model C1 – Average Environmental Game used as baseline b (95% CI) 1.10 (0.61, 1.59)*** -1.59 (-2.10, -1.08)*** -0.43 (-0.88, 0.01) -0.19 (-0.36, -0.03)* 0.21 (-0.07, 0.49) 0.22 (0.11, 0.32)*** -0.27 (-1.50, 0.97) 6.97 (3.06, 10.87)** 832 208 .11
Model C2 – Random Matching Game used as baseline b (95% CI) 1.54 (1.01, 2.07)*** -1.15 (-1.67, -0.63)*** 0.43 (-0.01, 0.88) -0.19 (-0.36, -0.03)* 0.21 (-0.07, 0.49) 0.22 (0.11, 0.32)*** -0.27 (-1.50, 0.97) 6.53 (2.60, 10.46)*** 832 208 .11
Notes: * p < .05, ** p < .01, *** p < .001, all tests are two-sided. Estimation method is robust linear regression with standard errors clustered at the subject level.
15
C3. Robustness (exploratory) analyses: Prediction of own contributions In Table C2 we repeat analyses reported in Table 3 in the paper, including in addition three control variables (risk preferences, age and gender). In Model C3, we regressed participants’ contributions to the Group project on their environmental score, on matching group type dummies (with the “Average Environmental Game” matching group serving as baseline) and the three control variables. Model C4 is the same as Model C3, but the “Random Matching Game” group serves as the baseline category. Mirroring the results reported in the main text, hypothesis H3 received support also in the robustness analyses and H4 did not. Older participants contributed more on average, but this result should be regarded as exploratory. Table C2 Regression of contributions on environmental score, matching group type and additional controls
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Environmental score Risk preferences Age Female Constant Observations No. of clusters R2
Model C3 – Average Environmental Game used as baseline b (95% CI) 1.22 (0.58, 1.87)*** -1.86 (-2.46, -1.25)*** -0.00 (-0.58, 0.57) 0.10 (-0.10, 0.29) -0.10 (-0.49, 0.28) 0.18 (0.06, 0.29)** 0.09 (-1.52, 1.70) 3.32 (-1.38, 8.02) 832 208 .06
Model C4 – Random Matching Game used as baseline b (95% CI) 1.23 (0.64, 1.81)*** -1.85 (-2.61, -1.09)*** 0.00 (-0.57, 0.58) 0.10 (-0.10, 0.29) -0.10 (-0.49, 0.28) 0.18 (0.06, 0.29)** 0.09 (-1.52, 1.70) 3.32 (-1.39, 8.02) 832 208 .06
Notes: ** p < .01, *** p < .001; all tests are two-sided. Estimation method is robust linear regression with standard errors clustered at the subject level.
C4. Exploratory analyses: Environmentalism as a predictor of cooperation To explore whether own environmental score can predict cooperation, depending on the level of one’s co-players’ environmentalism, we regressed own contribution to the Group project on one’s environmental score separately for each condition (see section 2.3.1). The results are reported in Table C3. Own environmentalism had no effect on contributions, irrespective of condition (and thus regardless of the environmentalism of one’s co-players). Table C3 Regression of contributions on environmental score
Environmental score Constant
Most Environmental Game b (95% CI) 0.19 (-0.05, 0.44) 6.69 (2.13, 11.24)**
Least Environmental Game b (95% CI) -0.01 (-0.23, 0.22) 7.20 (2.91, 11.49)**
16
Average Environmental Game b (95% CI) 0.12 (-0.09, 0.34) 6.74 (2.65, 10.82)**
Random Matching Game b (95% CI) 0.12 (-0.12, 0.36) 6.74 (2.29, 11.20)**
Observations R2
208 .01
208 .00
208 .01
208 .01
Notes: ** p < .01; all tests are two-sided. Estimation method is linear regression with robust standard errors.
17
Online Appendix D: Printscreens of instructions
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
One’s environmental behavior influences how others see and behave towards the actor. People who behave pro-environmentally are expected to be more cooperative. People who behave pro-environmentally are preferred as interaction partners. People who behave pro-environmentally elicit more cooperation from others. Results were obtained in incentivized decision tasks.
S0272-4944(19)30461-X
DOI:
https://doi.org/10.1016/j.jenvp.2019.101362
Reference:
YJEVP 101362
To appear in:
Journal of Environmental Psychology
Received Date: 16 July 2019 Revised Date:
7 October 2019
Accepted Date: 9 October 2019
Please cite this article as: Vesely, S., Klöckner, C.A., Brick, C., Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment, Journal of Environmental Psychology (2019), doi: https://doi.org/10.1016/j.jenvp.2019.101362. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment
Stepan Vesely, Norwegian University of Science and Technology, [email protected] Christian A. Klöckner, Norwegian University of Science and Technology, [email protected] Cameron Brick, Department of Psychology, University of Cambridge & Department of Psychology, Universiteit van Amsterdam, [email protected]
Pro-environmental behavior as a signal of cooperativeness: Evidence from a social dilemma experiment
Abstract Pro-environmental behavior has social signaling value. Previous research suggests that enacting pro-environmental behaviors can signal certain personal characteristics, such as social status and trustworthiness, to others. Using an incentivized experiment, we show that people known to behave pro-environmentally are expected to be more cooperative, are preferred as cooperation partners, and elicit more cooperation from others. The presence of pro-environmental individuals may thus motivate others to exert more effort towards reaching cooperative goals, even in situations where individual and group goals are at odds (i.e., social dilemmas). However, people who behaved pro-environmentally were actually no more cooperative than those performing fewer pro-environmental behaviors.
1. Introduction Climate change mitigation and many of the associated pro-environmental behaviors such as reducing air travel and meat consumption (Dietz, Gardner, Gilligan, Stern, & Vandenbergh, 2009) have become contentious and polarizing subjects in the public discourse (McCright & Dunlap, 2011). As a consequence, people may be reluctant to display or advocate mitigation actions in order to avoid conflict, social sanctions and being perceived negatively by others (cf. Bashir, Lockwood, Chasteen, Nadolny, & Noyes, 2013; Brick, Sherman, & Kim, 2017). Also, people may sort into social relationships with those sharing similar views regarding climate protection, which would allow them to exhibit (or not) pro-environmental behaviors more freely, but this assortative matching could lead to even greater opinion polarization in the long term (Böhm, Pfister, Salway, & Fløttum, 2019). One good strategy on how to start gaining more insight into the processes sketched in the previous paragraph is to examine how people behaving pro-environmentally (henceforth referred to as “environmentalists” for short) are perceived by others, and whether such perceptions can have real consequences in terms of subsequent social interaction. There is now a small but growing literature addressing the complex question of how environmentalists are perceived. What previous research shows is that pro-environmental behaviors and attitudes may be able to convey information about a person’s social status (Brooks & Wilson, 2015; Puska, Kurki, Lähdesmäki, Siltaoja, & Luomala, 2016; Sadalla & Krull, 1995; Skippon, Kinnear, Lloyd, & Stannard, 2016; but see Berger, 2017; Welte & Anastasio, 2010), trustworthiness (Fehrler & Kosfeld, 2013; but see Berger, 2017; Puska et al., 2016), and certain personality traits like conscientiousness, agreeableness, and altruism (Puska et al., 2016; Skippon & Garwood, 2011; Skippon et al., 2016). While existing research shows much promise, it has at least two limitations the present work attempts to address. First, we study not only whether behaving pro-environmentally shapes others’ perceptions of the actor, but also whether it can have an impact on subsequent social interaction. Second, we study whether people take others’ environmentalism into account when making decisions with real (financial) stakes on the line, as opposed to hypothetical decisions or survey responses (cf. Camerer & Hogarth, 1999; Kormos & Gifford, 2014; Klein, Hilbig, & Heck, 2017; Vesely & Klöckner, 2018a). 1
We extend the literature on the signaling function of environmentalism by testing whether pro-environmental behavior is perceived by others to signal cooperativeness, and by studying how people respond to such a signal. We also explore whether environmentalists behave more cooperatively themselves. Pro-environmental behavior can thus be linked to a key domain of social interaction (cooperation) through its function as a social signal. In this view, environmentally relevant decisions are not made in isolation from the broader social context, but drive and are driven by social influence processes. People may sometimes refrain from behaving pro-environmentally if they think this could harm their social image (Brick et al., 2017). However, if pro-environmental behavior can serve as a reliable signal of cooperativeness, applied research could then build on this finding to facilitate pro-environmental behavior. Specifically, environmentalists’ greater perceived cooperativeness and others’ greater willingness to cooperate with them could be leveraged as a co-benefit of and further motivation for behaving pro-environmentally. This possibility echoes what has been suggested by Keohane and Victor (2016) in the context of climate change mitigation negotiations: that countries may contribute to climate change mitigation to build their reputation and to secure future cooperation from others (see also Bain et al., 2016; Brekke & Nyborg, 2008). Thus, if pro-environmental behavior is able to signal cooperative tendencies, this can have important practical implications. Before proceeding to our hypotheses, we shall provide a brief outline of their key theoretical underpinnings. We study cooperation using a standard social dilemma setting (the public goods game) where the group’s material welfare is maximized by joint cooperation by everyone in the group, but at the same time an individual is materially better off not cooperating regardless of what the other group members do (Andreoni, 1988; Dawes, 1980). What motivates our prediction that people engaging in pro-environmental behavior are also more cooperative (see H4 below) is that both cooperative and pro-environmental behaviors are driven by pro-social preferences, such as inequity aversion and norm compliance in case of cooperation (Fehr & Schmidt, 1999; Kimbrough & Vostroknutov, 2016; Krupka & Weber, 2013) and attitudes, values, and norms in case of pro-environmental behavior (Bamberg & Möser, 2007; Klöckner, 2013). We specifically propose that pro-environmental and cooperative behaviors may be linked by virtue of their underlying pro-social motives being correlated within individuals. This is consistent with previous research linking cooperation to pro-environmental behaviors and attitudes (Kaiser & Byrka, 2011; Sussman, Lavallee, & Gifford, 2016; Tarditi, Hahnel, Jeanmonod, Sander, & Brosch, 2018; but see Smith & Bell, 1992), as well as with research on associations between different pro-social motives more generally (Bamberg & Möser, 2007; Gächter, Nosenzo, & Sefton, 2013; Kimbrough & Vostroknutov, 2016; Klöckner, 2013). We next propose that if environmentally-friendly individuals are indeed more cooperative outside the lab, observers will have formed a mental association between environmentalism and cooperation through experience (see H1). Because in social dilemmas, interaction with cooperators rather than free-riders is by design more advantageous (see section 2.3.1 and Andreoni, 1988), we expect that environmentalists, if perceived to be cooperative, will be preferred as cooperation partners (see H2). Finally, provided that environmentalists are believed to be cooperative, others should cooperate with them (see H3) due to well-established reciprocity and inequity aversion motives (Fehr & Schmidt, 1999). Hypotheses. H1: Individuals will expect those who perform more pro-environmental behaviors to be more cooperative in social dilemmas. H2: People who perform more pro2
environmental behaviors will be preferred as cooperation partners in social dilemmas. H3: People who perform relatively more pro-environmental behaviors will elicit more cooperation from others. H4: People who perform relatively more pro-environmental behaviors will be more cooperative in social dilemmas. To test our hypotheses, we invited participants to play four public goods games in the laboratory, select their preferred game partners, and respond to questionnaires, as detailed below. 2. Method 2.1 Participants and sessions Two hundred and eight participants (111 females; mean age = 23.94 years, SD = 5.45) recruited from a subject pool maintained by the Vienna Center for Experimental Economics (VCEE) took part in the study. A priori power calculations indicated that a sample of at least 191 participants was required to detect a small effect (partial R2 = .04) with alpha at .05 (twotailed) and statistical power at .80 (Faul, Erdfelder, Lang, & Buchner, 2007). ORSEE (Greiner, 2015) was used for recruitment and z-Tree (Fischbacher, 2007) for programming. Participants were compensated for their time, mean earnings = 35.3 EUR (including payment for additional unrelated tasks, see section B5 in Appendix B). The design was independently reviewed and approved by VCEE staff, in accordance with University of Vienna regulations. 2.2 Questionnaire Participants responded to 28 questions about their previous pro-environmental behaviors (most of the items were adapted from Kaiser, 1998; see Appendix D). We calculated the total score for each participant, with higher scores indicating enactment of more pro-environmental behaviors (M = 18.03, SD = 3.77, α = .70). Participants also answered socio-demographic questions. We then elicited participants’ risk preferences, using a task adapted from Eckel and Grossman (2002). This served as a filler task obscuring the link between the environmental behavior questionnaire and the social dilemma games. Importantly, we later gave participants the opportunity to condition their selection of partners in one of the social dilemma games on the potential partner’s risk preferences (see section 2.5). 2.3 Public goods games Participants played four standard one-shot linear public goods games (Andreoni, 1988) without feedback. In each game, participants formed anonymous groups of four. In each game, each participant was endowed with 20 tokens and had to decide how to distribute the tokens between a Group project and an Individual project. One token invested in the Individual project would yield one point to the investor and zero points to the other three group members. One token invested in the Group project would yield 0.5 points to all four group members. Thus, investing in the Group project (cooperation) maximized the group’s material welfare, while investing in the Individual project yielded higher material utility to the individual. At the end of the experiment, one of the four games was randomly selected for payment with an exchange rate 10 points = 2 EUR. 2.3.1 Group matching (treatments) In the “Most Environmental Game” condition, participants were informed they would be matched with the three most pro-environmental participants in the session (excluding 3
themselves) based on the pre-experimental questionnaire. In “Least Environmental Game”, participants were informed they would be matched with the three least pro-environmental participants in the session. In “Average Environmental Game”, participants were informed they would be matched with the three participants in the session that were closest to that session’s mean environmentalism score. In “Random Matching Game”, participants were informed they would be matched into groups randomly. The order of the four games was randomized across participants. 2.4 Belief elicitation After each game, participants guessed how many tokens others in their matching group contributed to the Group project in total in that game. Participants were given a 2 EUR bonus for guessing others’ total contribution correctly within ± 3 tokens. 2.5 Partner selection Participants next proceeded to play three repeated public goods games as a part of an unrelated study. Important for the present study, they were given an opportunity to select their preferred partners for one of these games from among other participants in the session. Specifically, each participant was shown a table listing all other participants in the session in random order, with the following information displayed in three columns: each participant’s environmental score, risk seeking score and age. The order in which the three types of scores were displayed in the columns was randomized across participants. Thus, participants were able to condition their choice of partners on two additional plausible criteria besides environmental score (i.e., others’ risk preferences and age, see Kocher, Martinsson, Matzat, & Wollbrant, 2015; Thöni, Tyran, & Wengström, 2012). Participants could then indicate on a 7point scale whether they would like to be matched with each of the other participants present. Participants knew that their stated preferences would with some probability determine with whom they would be matched (see Appendix B). 3. Results Figure 1 displays participants’ contributions to the Group project in the four games and their beliefs of how much a co-player in their matching group contributed.
4
Fig. 1: Own actual and co-player’s expected contributions (means and 95% CIs; maximum possible contribution per game was 20 tokens). In Table 1, we report estimates of two regression models testing H1. In Model 1, we regressed participants’ expectations of how much others in their matching group contributed to the Group project on matching group type dummies, with “Average Environmental Game” serving as the baseline category. Model 2 is the same as Model 1, but “Random Matching Game” serves as the baseline category. All reported tests are two-sided. Table 1 Regression of expected contributions on group type
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Constant Observations No. of clusters R2
Model 1 – Average Environmental Game used as baseline b (95% CI) 1.10 (0.62, 1.59)*** -1.59 (-2.09, -1.08)*** -0.43 (-0.88, 0.01) 9.41 (8.71, 10.12)*** 832 208 .03
Model 2 – Random Matching Game used as baseline b (95% CI) 1.54 (1.01, 2.07)*** -1.15 (-1.67, -0.63)*** 0.43 (-0.01, 0.88) 8.98 (8.29, 9.66)*** 832 208 .03
Notes: *** p < .001. Estimation method is robust linear regression with standard errors clustered at the subject level.
We found clear support for H1: Environmentalists were believed to be more cooperative. Specifically, compared to either control group (Average Environmental Game or Random Matching Game), participants expected co-players who were known to perform the most proenvironmental behaviors in one’s session to contribute more to the Group project, while the reverse was true for co-players who were known to perform the fewest pro-environmental behaviors. 5
The test of H2 is presented in Table 2. As described in section 2.5, for every individual in a session, each participant indicated whether they preferred to be matched with them on a 7point scale. For every individual we averaged the preference ratings received from others. Everyone was thus assigned a “mean preference rating”. Higher values indicated that others wanted to be matched with this person, and lower values indicated that others preferred not to be matched with this person. In Table 2, we regressed participants’ mean preference rating on their environmental score, risk seeking score and age score (i.e., the indicators displayed during the rating exercise). We found strong support for H2: Participants preferred to be matched with others who performed many pro-environmental behaviors. Age and risk preferences had no effect on being preferred as cooperation partner. Table 2 Regression of received preference ratings on players’ characteristics Environmental score Risk score Age score Constant Observations R2
b (95% CI) 0.07 (0.06, 0.08)*** 0.00 (-0.01, 0.01) 0.00 (-0.01, 0.00) 3.12 (2.93, 3.31)*** 208 .51
Notes: *** p < .001. Linear regression with robust standard errors.
Table 3 presents estimates of two regression models simultaneously testing H3 and H4. In Model 3, we regressed participants’ contributions to the Group project on their environmental score (to test H4) and on matching group type dummies (to test H3), with Average Environmental Game serving as the control group. Model 4 is the same as Model 3, but Random Matching Game serves as control. Table 3 Regression of contributions on environmental score and group type
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Environmental score Constant Observations No. of clusters R2
Model 3 – Average Environmental Game used as baseline b (95% CI) 1.22 (0.58, 1.86)*** -1.86 (-2.46, -1.25)*** -0.00 (-0.58, 0.57) 0.11 (-0.09, 0.30) 7.00 (3.33, 10.68)*** 832 208 .03
Model 4 – Random Matching Game used as baseline b (95% CI) 1.23 (0.64, 1.81)*** -1.85 (-2.61, -1.09)*** 0.00 (-0.57, 0.58) 0.11 (-0.09, 0.30) 7.00 (3.32, 10.68)*** 832 208 .03
Notes: *** p < .001. Robust linear regression with standard errors clustered at the subject level.
We found clear support for H3: Environmentalists elicited more cooperation from others. Specifically, compared to either control group, participants contributed more to the Group project when matched with co-players who were known to perform the most proenvironmental behaviors in one’s session, while they contributed less to the Group project when matched with co-players who were known to perform the fewest pro-environmental behaviors. Hypothesis H4 was not supported: Those who reported performing more pro6
environmental behaviors did not cooperate more than those performing fewer proenvironmental behaviors. 4. Discussion and conclusions In this paper we focused on the signaling value of pro-environmental behavior and specifically on its ability to signal cooperative tendencies. We posited that pro-environmental behavior can serve as a signal of cooperation (H1), and that others would act on this signal (H2, H3), and we also proposed that environmentalists would themselves be more cooperative (H4). As predicted, environmentalists were perceived as more cooperative in the social dilemma setting (H1), they were sought after as cooperation partners (H2), and they elicited more cooperation from others (H3). Unexpectedly, environmentalists were not more cooperative compared to those performing fewer pro-environmental behaviors (H4). These findings contribute to the literature on the signaling value of environmentalism (e.g., Berger, 2017; Brick et al., 2017; Griskevicius, Tybur, & Van den Bergh, 2010) by showing that environmental behaviors signal a cooperative disposition and that others act on this signal in situations with real consequences. An important implication of our results is that being seen as an environmentalist confers personal side-benefits to environmentalists, as others are more cooperative towards such individuals, and more willing to establish interactions with them (a form of social capital, see Cinyabuguma, Page, & Putterman, 2005). Such side-benefits could be leveraged by policymakers to promote further pro-environmental action (Bain et al., 2016; Griskevicius et al., 2010; Keohane & Victor, 2016). Environmentalists, however, do not appear to fulfill others’ expectations of being more cooperative themselves (see H4). Subsequent research should therefore explore how to motivate environmentalists to meet others’ initial positive expectations (see e.g. Bellemare, Sebald, & Suetens, 2018). Otherwise, cooperation is likely to break down in repeated encounters when initial positive expectations are not met (Fischbacher & Gächter, 2010). A possible explanation for failing to support H4 could be that cooperative decisions were made anonymously in our experiment, while past cooperative decisions involved in expectation formation had to be publicly observable. Decision observability promotes pro-social behavior (Bradley, Lawrence, & Ferguson, 2018), which may help explain why anonymous decisions deviate from others’ expectations. An intriguing hypothesis consistent with our data is that environmentalists are particularly sensitive to being publicly observed. This suggests possible extensions of the present study, linking it to research on the role that decision observability plays in environmental behavior (Griskevicius et al., 2010; Brick et al., 2017; Vesely & Klöckner, 2018b). A limitation of our study is that environmental behaviors were self-reported, rather than objectively measured (Kormos & Gifford, 2014; Lange & Dewitte, 2019). Another limitation is that we relied on a single decision paradigm, the “give some” public goods game, and replications using other types of social dilemmas could point to possible boundary conditions of the reported effects. There are for example subtle behavioral differences between “give some” and “take some” social dilemmas (Fosgaard, Hansen, & Wengström, 2014). One could speculate that different types of pro-environmental behavior (e.g., investment vs. curtailment) could be better able to signal cooperativeness in specific types of dilemmas (such as the give some vs. take some dilemmas). Another limitation is that only two “distractor” characteristics (age and risk preferences) were used along with environmentalism to inform participants about their potential cooperation partners. Finally, we cannot altogether rule out experimenter 7
demand, as information about others’ pro-environmental behavior was somewhat prominent in the decision tasks and some participants may have felt compelled to react to it for this reason.
References Andreoni, J. (1988). Why free ride? Strategies and learning in public goods experiments. Journal of Public Economics, 37, 291-304. Bain, P. G., Milfont, T. L., Kashima, Y., Bilewicz, M., Doron, G., Garðarsdóttir, R. B., et al. (2016). Co-benefits of addressing climate change can motivate action around the world. Nature Climate Change, 6, 154-157. Bamberg, S., & Möser, G. (2007). Twenty years after Hines, Hungerford, and Tomera: A new meta-analysis of psycho-social determinants of pro-environmental behaviour. Journal of Environmental Psychology, 27, 14-25. Bashir, N. Y., Lockwood, P., Chasteen, A. L., Nadolny, D., & Noyes, I. (2013). The ironic impact of activists: Negative stereotypes reduce social change influence. European Journal of Social Psychology, 43, 614-626. Bellemare, C., Sebald, A., & Suetens, S. (2018). Heterogeneous guilt sensitivities and incentive effects. Experimental Economics, 21, 316-336. Berger, J. (2017). Are luxury brand labels and “green” labels costly signals of social status? An extended replication. PLoS ONE, 12, e0170216. Böhm, G., Pfister, H.-R., Salway, A. & Fløttum, K. (2019). Remembering and communicating climate change narratives: The influence of world views on selective recollection. Frontiers in Psychology, 10, article 1026. Brekke, K. A., & Nyborg, K. (2008). Attracting responsible employees: Green production as labor market screening. Resource and Energy Economics, 30, 509-526. Brick, C., Sherman, D. K., & Kim, H. S. (2017). “Green to be seen” and “brown to keep down”: Visibility moderates the effect of identity on pro-environmental behavior. Journal of Environmental Psychology, 51, 226-238. Brooks, J. S., & Wilson, C. (2015). The influence of contextual cues on the perceived status of consumption-reducing behavior. Ecological Economics, 117, 108-117. Camerer, C., & Hogarth, R. M. (1999). The effects of financial incentives in experiments: A review and capital-labor-production framework. Journal of Risk and Uncertainty, 19, 7-42. Cinyabuguma, M., Page, T., & Putterman, L. (2005). Cooperation under the threat of expulsion in a public goods experiment. Journal of Public Economics, 89, 1421-1435. Dawes, R. M. (1980). Social dilemmas. Annual Review of Psychology, 31, 169-193.
8
Dietz, T., Gardner, G. T., Gilligan, J., Stern, P. C. & Vandenbergh, M. P. (2009). Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Proceedings of the National Academy of Sciences, 106, 18452-18456. Eckel, C., & Grossman, P. J. (2002). Sex differences and statistical stereotyping in attitudes toward financial risk. Evolution and Human Behavior, 23, 281-295. Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175-191. Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. The Quarterly Journal of Economics, 114, 817-868. Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10, 171-178. Fischbacher, U., & Gächter, S. (2010). Social preferences, beliefs, and the dynamics of free riding in public good experiments. American Economic Review, 100, 541-556. Fosgaard, T. R., Hansen, L. G., & Wengström, E. (2014). Understanding the nature of cooperation variability. Journal of Public Economics, 120, 134-143. Gächter, S., Nosenzo, D., & Sefton, M. (2013). Peer effects in pro-social behavior: Social norms or social preferences? Journal of the European Economic Association, 11, 548-573. Greiner, B. (2015). Subject pool recruitment procedures: Organizing experiments with ORSEE. Journal of the Economic Science Association, 1, 114-125. Griskevicius, V., Tybur, J. M., & Van den Bergh, B. (2010). Going green to be seen: Status, reputation, and conspicuous conservation. Journal of Personality and Social Psychology, 98, 392-404. Kaiser, F. G. (1998). A general measure of ecological behavior. Journal of Applied Social Psychology, 28, 395-422. Kaiser, F. G., & Byrka, K. (2011). Environmentalism as a trait: Gauging people’s prosocial personality in terms of environmental engagement. International Journal of Psychology, 46, 71-79. Keohane, R. O., & Victor, D. G. (2016). Cooperation and discord in global climate policy. Nature Climate Change, 6, 570-575. Kimbrough, E. O., & Vostroknutov, A. (2016). Norms make preferences social. Journal of the European Economic Association, 14, 603-638. Klein, S. A., Hilbig, B. E., & Heck, D. W. (2017). Which is the greater good? A social dilemma paradigm disentangling environmentalism and cooperation. Journal of Environmental Psychology, 53, 40-49.
9
Klöckner, C. A. (2013). A comprehensive model of the psychology of environmental behavior: A meta-analysis. Global Environmental Change, 23, 1028-1038. Kocher, M. G., Martinsson, P., Matzat, D., & Wollbrant, C. (2015). The role of beliefs, trust, and risk in contributions to a public good. Journal of Economic Psychology, 51, 236-244. Kormos, C., & Gifford, R. (2014). The validity of self-report measures of proenvironmental behavior: A meta-analytic review. Journal of Environmental Psychology, 40, 359-371. Krupka, E. L., & Weber, R. A. (2013). Identifying social norms using coordination games: Why does dictator game sharing vary? Journal of the European Economic Association, 11, 495-524. Lange, F., & Dewitte, S. (2019). Measuring pro-environmental behavior: Review and recommendations. Journal of Environmental Psychology, 63, 92-100. McCright, A. M., & Dunlap, R. E. (2011). The politicization of climate change and polarization in the American public's views of global warming, 2001-2010. The Sociological Quarterly, 52, 155-194. Puska, P., Kurki, S., Lähdesmäki, M., Siltaoja, M., & Luomala, H. (2016). Male-male status signaling through favoring organic foods: Is the signaler perceived and treated as a friend or a foe? Psychology & Marketing, 33, 843-855. Sadalla, E. K., & Krull, J. L. (1995). Self-presentational barriers to resource conservation. Environment and Behavior, 27, 328-353. Skippon, S., & Garwood, M. (2011). Responses to battery electric vehicles: UK consumer attitudes and attributions of symbolic meaning following direct experience to reduce psychological distance. Transportation Research Part D, 16, 525-531. Skippon, S. M., Kinnear, N., Lloyd, L., & Stannard, J. (2016). How experience of use influences mass-market drivers’ willingness to consider a battery electric vehicle: A randomised controlled trial. Transportation Research Part A, 92, 26-42. Smith, J. M., & Bell, P. A. (1992). Environmental concern and cooperative-competitive behavior in a simulated commons dilemma. The Journal of Social Psychology, 132, 461-468. Sussman, R., Lavallee, L. F., & Gifford, R. (2016). Pro-environmental values matter in competitive but not cooperative commons dilemmas. The Journal of Social Psychology, 156, 43-55. Tarditi, C., Hahnel, U. J. J., Jeanmonod, N., Sander, D., & Brosch, T. (2018). Affective dilemmas: The impact of trait affect and state emotion on sustainable consumption decisions in a social dilemma task. Environment and Behavior. Thöni, C., Tyran, J.-R., & Wengström, E. (2012). Microfoundations of social capital. Journal of Public Economics, 96, 635-643.
10
van der Werff, E., Steg, L., & Keizer, K. (2014). Follow the signal: When past proenvironmental actions signal who you are. Journal of Environmental Psychology, 40, 273282. Vesely, S., & Klöckner, C. A. (2018a). Global social norms and environmental behavior. Environment and Behavior, 50, 247-272. Vesely, S., & Klöckner, C. A. (2018b). How anonymity and norms influence costly support for environmental causes. Journal of Environmental Psychology, 58, 27-30. Welte, T. H. L., & Anastasio, P. A. (2010). To conserve or not to conserve: Is status the question? Environment and Behavior, 42, 845-863.
11
Online Appendix A: Descriptive statistics Table A1 presents key descriptive statistics, namely means and SDs (in brackets) summarizing participants’ contributions to the Group project in the four games and their beliefs of how much each of the co-players in their matching group contributed on average. Raw data are available upon request. Table A1 Summary of contributions and beliefs (means, SDs)
Own contribution to the Group project Expected average contribution to the Group project by a co-player
Most Environmental Game 10.16 (6.60)
Least Environmental Game 7.08 (6.24)
Average Environmental Game 8.94 (6.18)
Random Matching Game 8.93 (6.44)
10.52 (5.41)
7.83 (5.10)
9.41 (5.17)
8.98 (5.01)
12
Online Appendix B: Further details on methods B1. Risk elicitation task (see section 2.2) To elicit risk preferences, we used a task where one had to choose one of eight incentivized lotteries (played for real money) involving various degrees of risk (Eckel & Grossman, 2002). An example of a low risk lottery was a prospect of getting either 4.2 EUR or 6.0 EUR with equal probability, whereas an example of a high risk lottery was a prospect of getting either 1.4 EUR or 10.8 EUR with equal probability. Participants were assigned a risk seeking score which was equal to 1 if the safest lottery was selected and equal to 8 if the most risky lottery was selected, with values in between for lotteries involving intermediate levels of risk. B2. Recoding responses when calculating scale reliability (see section 2.2) When calculating scale reliability for the 28-item environmental behavior questionnaire, all values were recoded so that higher values indicated more pro-environmental behavior. We also did the following changes: For item “Are you de-icing the refrigerator regularly?”, the response “I share the refridgerator with other people and someone else is responsible for this” was coded as missing, as it did not unambiguously reflect a pro- or anti-environmental behavior. We acknowledge that this coding decision (and similar coding decisions below) reflect a judgment call on our part. For item “Are you getting your electricity from a provider that guarantees 100% regenerative energy production?”, the response “I can not influence at all from which provider to get electricity” was coded as missing. For item “Do you use a tumble dryer to dry clothes?”, the response “I do not have a tumble dryer” was recoded as “No”. For item “Do you bring your own coffee (tea) cup to school/work (instead of using cups from the coffee machine)?”, the response “I do not drink coffee (tea) at school/work” was coded as missing. For item “Do you drive a car with low fuel consumption or an alternative fuel car (e.g., an electric vehicle)?”, the response “I do not drive” was assigned the same value as “Yes” (i.e., both being coded as the maximally pro-environmental response). For item “Do you drive your car defensively to safe fuel?”, the response “I do not drive” was assigned the same value as “Yes” (i.e., both being coded as the maximally pro-environmental response). For item “Do you return dead batteries to collection points for dangerous waste?”, the response “I do not use batteries” was coded as missing. For item “Do you buy organic meat products?”, the response “I do not eat dairy products” was coded as missing. For item “Do you buy organic meat products?”, the response “I do not eat meat products” was assigned the same value as “Yes” (i.e., both being coded as the maximally pro-environmental response).
13
B3. Comprehension check (see section 2.3) Before proceeding to actual game play, participants had to correctly answer control questions checking their comprehension of the game, and were given an opportunity to try out different investment strategies using an interface automatically calculating outcomes for each selected strategy. This was done in order to ensure that participants fully understood the social dilemma structure of the decision situation. B4. Partner selection: Details of the matching protocol (see section 2.5) Here we describe how we ensured that the preferences for cooperation partners elicited from participants in the task described in section 2.5 in the paper were, with some probability, able to determine with whom participants were subsequently matched. Technically, after participants submitted their preferences, the computer randomly assigned a “position” to every participant between 1 and n, where n was the number of participants in the session. The matching then proceeded in the following way: First, the participant with the highest position was matched with his or her three most preferred co-players (with ties broken at random). Second, the participant with the highest position among the un-matched participants was matched with his or her three most preferred co-players among those that have not yet been matched (again with ties broken at random), and so on until everyone was matched. Participants were made aware of this procedure beforehand. B5. Unrelated tasks conducted in the session (not reported here) Participants also played an additional one-shot public goods game and three iterated public goods games at the end of the session.
14
Online Appendix C: Additional analyses C1. Non-parametric tests of H1 and H3 Results of Wilcoxon signed-rank tests lead to the same conclusion as the regression tests of H1 reported in Table 1 in the paper. Specifically, expected contributions by matching group members were higher in Most Environmental Game than in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). Expected contributions by matching group members in Least Environmental Game were, as predicted, lower than in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). Results of Wilcoxon signed-rank tests similarly lead to the same conclusion as the regression tests of H3 reported in Table 3 in the main text. Contributions to the Group project were higher in Most Environmental Game than contributions in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). Contributions in Least Environmental Game were, again as predicted, lower than in Average Environmental Game (p < .001) and in Random Matching Game (p < .001). C2. Robustness (exploratory) analyses: Prediction of expected contributions In Table C1 we repeat analyses reported in Table 1 in the paper, with additional controls (own environmental score, risk preferences, age and gender) added to the models. In Model C1, we regressed participants’ expectations of how much others in their matching group contributed to the Group project on matching group type dummies and the additional control variables. “Average Environmental Game” matching group served as the baseline category. Model C2 is the same as Model C1, but the “Random Matching Game” group served as the baseline category. As in the main analyses, hypothesis H1 received support also in the robustness analyses (in fact, matching group dummy coefficients remain virtually unchanged). Interestingly, individuals performing more pro-environmental behaviors expected others to contribute less. Older participants expected others to contribute more. The latter results should be interpreted with caution, as these analyses are exploratory in nature. Table C1 Regression of expected contributions on matching group type and additional controls
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Environmental score Risk preferences Age Female Constant Observations No. of clusters R2
Model C1 – Average Environmental Game used as baseline b (95% CI) 1.10 (0.61, 1.59)*** -1.59 (-2.10, -1.08)*** -0.43 (-0.88, 0.01) -0.19 (-0.36, -0.03)* 0.21 (-0.07, 0.49) 0.22 (0.11, 0.32)*** -0.27 (-1.50, 0.97) 6.97 (3.06, 10.87)** 832 208 .11
Model C2 – Random Matching Game used as baseline b (95% CI) 1.54 (1.01, 2.07)*** -1.15 (-1.67, -0.63)*** 0.43 (-0.01, 0.88) -0.19 (-0.36, -0.03)* 0.21 (-0.07, 0.49) 0.22 (0.11, 0.32)*** -0.27 (-1.50, 0.97) 6.53 (2.60, 10.46)*** 832 208 .11
Notes: * p < .05, ** p < .01, *** p < .001, all tests are two-sided. Estimation method is robust linear regression with standard errors clustered at the subject level.
15
C3. Robustness (exploratory) analyses: Prediction of own contributions In Table C2 we repeat analyses reported in Table 3 in the paper, including in addition three control variables (risk preferences, age and gender). In Model C3, we regressed participants’ contributions to the Group project on their environmental score, on matching group type dummies (with the “Average Environmental Game” matching group serving as baseline) and the three control variables. Model C4 is the same as Model C3, but the “Random Matching Game” group serves as the baseline category. Mirroring the results reported in the main text, hypothesis H3 received support also in the robustness analyses and H4 did not. Older participants contributed more on average, but this result should be regarded as exploratory. Table C2 Regression of contributions on environmental score, matching group type and additional controls
Most Environmental Game Least Environmental Game Average Environmental Game Random Matching Game Environmental score Risk preferences Age Female Constant Observations No. of clusters R2
Model C3 – Average Environmental Game used as baseline b (95% CI) 1.22 (0.58, 1.87)*** -1.86 (-2.46, -1.25)*** -0.00 (-0.58, 0.57) 0.10 (-0.10, 0.29) -0.10 (-0.49, 0.28) 0.18 (0.06, 0.29)** 0.09 (-1.52, 1.70) 3.32 (-1.38, 8.02) 832 208 .06
Model C4 – Random Matching Game used as baseline b (95% CI) 1.23 (0.64, 1.81)*** -1.85 (-2.61, -1.09)*** 0.00 (-0.57, 0.58) 0.10 (-0.10, 0.29) -0.10 (-0.49, 0.28) 0.18 (0.06, 0.29)** 0.09 (-1.52, 1.70) 3.32 (-1.39, 8.02) 832 208 .06
Notes: ** p < .01, *** p < .001; all tests are two-sided. Estimation method is robust linear regression with standard errors clustered at the subject level.
C4. Exploratory analyses: Environmentalism as a predictor of cooperation To explore whether own environmental score can predict cooperation, depending on the level of one’s co-players’ environmentalism, we regressed own contribution to the Group project on one’s environmental score separately for each condition (see section 2.3.1). The results are reported in Table C3. Own environmentalism had no effect on contributions, irrespective of condition (and thus regardless of the environmentalism of one’s co-players). Table C3 Regression of contributions on environmental score
Environmental score Constant
Most Environmental Game b (95% CI) 0.19 (-0.05, 0.44) 6.69 (2.13, 11.24)**
Least Environmental Game b (95% CI) -0.01 (-0.23, 0.22) 7.20 (2.91, 11.49)**
16
Average Environmental Game b (95% CI) 0.12 (-0.09, 0.34) 6.74 (2.65, 10.82)**
Random Matching Game b (95% CI) 0.12 (-0.12, 0.36) 6.74 (2.29, 11.20)**
Observations R2
208 .01
208 .00
208 .01
208 .01
Notes: ** p < .01; all tests are two-sided. Estimation method is linear regression with robust standard errors.
17
Online Appendix D: Printscreens of instructions
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
One’s environmental behavior influences how others see and behave towards the actor. People who behave pro-environmentally are expected to be more cooperative. People who behave pro-environmentally are preferred as interaction partners. People who behave pro-environmentally elicit more cooperation from others. Results were obtained in incentivized decision tasks.